Clinical Factors Associated With Failed Weaning From Intraoperative Extracorporeal Membrane Oxygenation Used During Lung Transplantation

Background: Extracorporeal membrane oxygenation (ECMO) promotes adequate oxygenation and hemodynamic stability during lung transplantation (LTx). However, some recipients cannot be weaned from ECMO following surgery. Thus, we evaluated the prognosis and risk factors of failed weaning from intraoperative ECMO during LTx. Methods: We retrospectively analyzed data from 274 patients receiving intraoperative ECMO during LTx. Risk factors were evaluated using logistic regression analyses. Results: Weaning failure occurred in 118 patients (43.1%). Intensive care unit stay was longer and mortality was higher in the failed weaning group than in the successful weaning group. The failed weaning group exhibited signicantly older donor age, lower donor PaO 2 /FiO 2 ratio, greater intraoperative transfusion volume, and longer operation time than the successful weaning group. Recipient age, body mass index, donor age, lower donor PaO 2 /FiO 2 ratio, donor/recipient total lung capacity (TLC) ratio, greater intraoperative transfusion volume, and longer operation time were associated with weaning failure after adjustment. Conclusion: The failed weaning group showed a poor prognosis. Perioperative factors including donor age, donor PaO 2 /FiO 2 ratio, donor/recipient TLC, operation time, and blood loss can predict postoperative ECMO weaning failure.

revealed that the mortality risk was higher among patients receiving ECMO support after LTx than among those not receiving support; however, there was no difference in survival beyond 1 year between the two groups [16].
Several studies have analyzed risk factors and survival rates among patients who underwent LTx with planned postoperative extended ECMO due to severe pulmonary hypertension (PH) [9,10,17]. To the best of our knowledge, few studies have focused on factors associated with prolonged intraoperative ECMO after LTx [1,18]. Therefore, we aimed to investigate the prognosis of patients who continued to receive ECMO support following LTx and the risk factors for weaning failure from intraoperative ECMO in these patients.

Study design and population
In this single-center retrospective study, we reviewed data from consecutive patients who underwent LTx at Severance Hospital in South Korea from October 2012 to September 2020. Patients who underwent multi-organ transplantation (lung-liver, n=2; lung-kidney, n=2; heart-lung, n=1), those who did not receive intraoperative ECMO (CPB, n=6), those who underwent re-transplantation (n=5), and those aged ≤18 years (n=4) were excluded from the study. The remaining 274 patients were divided into two groups based on the success or failure of weaning from intraoperative ECMO immediately following LTx.
All LTx procedures were performed under ECMO support. Most patients received femoro-femoral venoarterial ECMO unless bridging ECMO was required during the waiting period. Weaning from ECMO was attempted post-surgery, following which patients were transferred to the ICU. The decision to wean from ECMO was determined according to the patient's status after reducing ECMO ow to 0.5 L/min. Under conditions of hemodynamic instability, such as (a) the need for a high-dose vasopressor, (b) decreased cardiac function based on transesophageal echocardiography, or (c) systolic blood pressure <100 mmHg despite treatment with norepinephrine (>0.2 µg/kg/min) and vasopressin (>0.05 U/min), venoarterial ECMO was maintained. Furthermore, ECMO support (veno-venous) was maintained despite hemodynamic stability if PaO 2 /FiO 2 was <150 mmHg with a positive end-expiratory pressure of 8 cmH 2 O.
Conversely, ECMO support was removed when hemodynamic stability and the target saturation were achieved.
All patients underwent induction immunosuppression therapy with high-dose corticosteroids (methylprednisolone at 250 mg during the operation and 0.5 mg/kg/day for 3 days after the operation). Triple immunosuppression therapy (e.g., prednisolone, tacrolimus, and mycophenolate mofetil) was used to maintain immunosuppression after transplantation. Patients who received bridging ECMO from January 2019 onward received 20 mg of basiliximab during transplantation surgery and were initiated on tacrolimus after 7 days. Ganciclovir and itraconazole were used in all recipients until 6 months postoperatively. Lifelong trimethoprim/sulfamethoxazole was used in all recipients to prevent Pneumocystis jirovecii after LTx.
Data collection for determining risk factors of failed weaning from intraoperative ECMO All data were collected from the electronic medical records of the hospital. Baseline data prior to LTx including demographic characteristics, cardiac function, comorbidities, and status while waiting were collected. Operative data included operation time, input and output of uid and blood, and ischemic time in the donor lung. Additionally, data related to the donor's demographics, PaO 2 /FiO 2 on the day of donation, and the time of mechanical ventilator application in the donor lung were collected.

Ethical approval
This research protocol was approved by the Institutional Review Board of Severance Hospital, South Korea (IRB No. 4-2021-0199), and the study design was approved by the appropriate ethics review boards. The requirement for obtaining patient informed consent was waived due to the retrospective nature of the study.

Statistical analysis
All statistical analyses were performed using IBM SPSS version 25.0 (IBM Corporation, Armonk, NY).
Continuous data are expressed as mean±standard deviation, and categorical data are expressed as numbers with corresponding percentages. Continuous and categorical variables were analyzed using Student's t-tests and chi-square/Fisher's exact tests, respectively.
A multivariate logistic regression model was used to identify independent risk factors for prolonged intraoperative ECMO. The model included variables with a level of signi cance <0.05 in the bivariate analysis. Odds ratios (ORs) and 95% con dence intervals (CIs) were calculated. For some continuous data, receiver operating characteristic (ROC) analysis was used to determine the cut-off value using the area under the curve (AUC). Survival data were estimated using the Kaplan-Meier method, and signi cant differences were determined using the log-rank test. Statistical signi cance was set at p<0.05.

Study population
During the study period, 294 patients underwent LTx. Of these, 274 were enrolled for analysis. The mean patient age was 54.6±11.4 years, and 63.9% of patients (n=175) were male. The major reason for LTx was idiopathic pulmonary brosis (n=149, 54.4%), followed by connective tissue disease-associated interstitial lung disease (n=49, 17.9%). During the LTx waiting period, 65.7% (n=180) of the enrolled patients were in the ICU, and 31% (n=85) received bridging ECMO. Immediately following the operation, 118 patients (43.1%) were admitted to the ICU while maintaining ECMO (Supplementary Table 1).
Comparison between the successful and failed weaning groups after LTx Among the enrolled patients, 118 (43.1%) were not weaned from ECMO, while 156 (56.9%) were successfully weaned from ECMO after the operation. Table 1 shows the results of the comparison between the two groups. The proportion of female patients and body mass index (BMI) were signi cantly higher in the failed weaning group than in the successful weaning group (female sex, 43.2% vs. 30.8%, p=0.034; BMI, 21.7±4.1 vs. 20.4±4.0 kg/m 2 , p=0.009). Perioperatively, the failed weaning group exhibited longer operation times, a larger amount of blood loss, and higher uid intake and transfusion volumes than the successful weaning group (mean operation time, 513.9 vs. 479.8 min, p=0.001; blood loss, 3.4 vs. 2.7 L, p=0.030; uid intake, 12.1 vs. 10.2 L, p=0.011, transfusion volume, 3.4 vs. 2.7 L, p=0.027). Among donor-related variables, there were signi cant differences in age, PaO 2 /FiO 2 ratio, and predicted donor/recipient total lung capacity (TLC) ratio between the two groups. Donors were signi cantly older in the failed weaning group than in the successful weaning group (44.7 vs. 41.9 years, p=0.014). The PaO 2 /FiO 2 ratio of the donor lung was signi cantly lower in the failed weaning group than in the successful weaning group (432.6 vs. 472.8, p<0.001), while the predicted donor/recipient TLC was higher (110.1 vs 105.2%, p=0.034).

Prognosis according to the success of intraoperative ECMO weaning
The failed weaning group exhibited a signi cantly longer ICU stay and duration of hospitalization after LTx than the successful weaning group (length of ICU stay, 24.5 vs. 9.0 days, p<0.001; length of hospitalization: 82.5 vs. 63.6 d, p=0.023, respectively). The mortality rates at 6 months and 1 year were signi cantly higher in the failed weaning group than in the successful weaning group (6 months, 29.7 vs 17.9%, p=0.023, 1 year, 43.2 vs. 26.8%, p=0.005). An analysis of overall survival during the observation period (October 2012 to May 2021) revealed that mortality rates were higher in the failed weaning group than in the successful weaning group (p=0.002, Figure 1).   (Table 3; Supplementary table 3). The additional analysis revealed that the duration of ICU stay and hospitalization were longer and the mortality rates at 6 months and 1 year were signi cantly higher in the failed weaning group than in the successful weaning group. Univariate analysis revealed signi cant differences in sex, total uid intake and transfusion volume during the operation, donor age, and predicted donor/recipient TLC between the two groups. A multivariate analysis including variables identi ed as signi cant in the univariate analysis revealed that BMI, transfusion volume >3.8 L, and PaO 2 /FiO 2 ratio in the donor lung were independent risk factors for intraoperative ECMO weaning failure among patients receiving bridging ECMO while waiting for LTx.

Discussion
Herein, we investigated the risk factors associated with failed weaning from intraoperative ECMO and prognosis among patients undergoing LTx. Our ndings indicated that hospitalization periods were longer and survival rates were lower in the failed weaning group than in the successful weaning group. Preoperative factors that increased the risk of weaning failure included older age of the donor, lower PaO 2 /FiO 2 ratio of the donor lung, and higher predicted donor/recipient TLC ratio. Intraoperative factors in uencing weaning failure after LTx included the duration of operation and transfusion volume. Similar results were obtained in the subgroup analysis of patients who underwent preoperative bridging ECMO.
Conventionally, our center has used ECMO during LTx since March 2013 [6]. The routine use of intraoperative ECMO during LTx allows for controlled perfusion and protective ventilation of the graft during the procedure, thus reducing the risk of later primary graft dysfunction (PGD) [6, 10]. However, extended ECMO is sometimes required in the immediate postoperative period. Several studies have analyzed outcomes among patients with planned continuation of intraoperative ECMO into the postoperative period, which is performed to allow adaptation of the left ventricle to the new loading conditions in patients with severe PH after LTx [9][10][11]19]. Results for these patients did not differ from those for patients without severe PH. However, Dell' Amore et al. reported a lower incidence of PGD and improved survival in the planned prolonged ECMO group [12]. While prolonged ECMO does not necessarily indicate a poor prognosis, unintended prolonged ECMO may be a marker of recipient status in terms of early graft failure and hemodynamic status [17]. Although several sequential studies have examined survival outcomes after postoperative ECMO, their outcomes have varied [14,[20][21][22]. Indeed, no studies have identi ed the preoperative and intraoperative factors associated with the need for extended ECMO following LTx. In our center, Narm et al. analyzed data from 74 patients after LTx. Among them, patients maintained on ECMO for >5 days after surgery exhibited higher mortality during the rst year than those maintained on ECMO for <4 days [18]. The authors also noted that increasing donor age, donor PaO 2 , and increasing operation time were independent risk factors for ECMO weaning failure after surgery [18].
Our analysis indicated that donor age was signi cantly correlated with ECMO weaning after surgery. Several studies have also reported that advanced donor age is an independent risk factor for extended ECMO [18,20]. Theoretically, older lungs may exhibit increased susceptibility to infection and reduced lung function [23]. While recipient age is a well-known prognostic factor for LTx, it did not signi cantly affect the success of ECMO weaning in our study [24].
Donor PaO 2 /FiO 2 ratio is associated with early gas exchange in the recipient [25]. Donor PaO 2 /FiO 2 was also a signi cant risk factor for failed ECMO weaning after surgery. However, there were no differences in allograft ischemic time between the two groups; this factor did not in uence the success of ECMO weaning. Although the correlation between ischemic time and pulmonary function or survival remains controversial [26,27], prolonged graft ischemic time-in older-aged donors-can lead to an adverse interaction [23].
Donor criteria do not include the duration of mechanical ventilation; however, infection in the donor lung has always been considered an absolute contraindication for LTx [25,28]. Of course, length of intubation is associated with bronchial colonization and predisposes the patient to ventilation-acquired pneumonia [25,28]. There were no differences in mechanical ventilation between our two study groups, and the duration of ventilation was approximately 7 days.
Surgical variables-long operation time, substantial blood loss due to severe pleural adhesion, delayed harvest team arrival, lengthy hemostasis, and unexpected anatomical or technical di culties-are adversely related to ECMO weaning. These factors may lead to lower postoperative oxygenation and aggravate pulmonary edema. Geube et al. noted an association between transfusion of a higher volume of red blood cells and the development of grade-3 PGD [29].
Historically, size matching has been considered important in LTx [23]. Size mismatch may in uence LTx outcomes, and several studies have demonstrated that there are no clinical or functional adverse effects when the donor predicted TCL is between 75% and 125% of the predicted value for the recipient [23]. Here, grafts were larger in the failed weaning group than in the successful weaning group, although this did not signi cantly in uence ECMO weaning.
Patients with severe PH often exhibit signi cant right ventricular dysfunction, decreased cardiac output, and hemodynamic instability [9]. As in other studies, extended ECMO was more likely to be required in these patients. However, there were no differences in mean pulmonary artery pressure or cardiac function between the two groups in our study (Table 1; Supplementary Table 4). The requirement for obtaining patient informed consent was waived due to the retrospective nature of the study.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request

Competing interests
The authors declare that they have no competing interests. Comparison of overall survival between the failed and successful extracorporeal membrane oxygenation (ECMO) weaning groups. Failed ECMO weaning was signi cantly associated with poor survival (log-rank p-value=0.002).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementarytable.docx